BBI Blog

Modernizing for AI: Start with the problem, not the platform

Written by BBI Leadership | Nov 18, 2026, 8:01:53 PM

Your company can make smarter cloud and AI decisions without overspending or creating unnecessary complexity. Yes, really.

Every technology vendor out there has a compelling reason why your organization needs a new platform, a lakehouse, or an AI agent. But before committing to yet another technology investment, business leaders need to answer a fundamental question: what problem are we actually trying to solve?

For organizations that delayed cloud modernization because the financial case wasn’t compelling, AI has brought the conversation back to the forefront.

In the first episode of BBI’s The Road to AI webinar series, The Cloud Shopper’s Guide, BBI experts explored how organizations can approach modernization more deliberately, preparing for AI without launching costly transformation programs even before the value is clear.

Here are the tactical tips our experts offered for ensuring your next technology investment is based on solving real business problems, not just flashy platforms.

Start with the business problem before choosing an AI or cloud platform

One of the most common mistakes in technology investment is choosing a platform before defining the problem it needs to solve. A better approach is to identify a specific business challenge plaguing your organization, such as time-consuming information searches, inconsistent processes, or limited access to useful insights.

When your problem is actually clear, it becomes easier to evaluate your technology options. Not every organization needs the same architecture – no matter what vendors might tell you – and a lakehouse or certain cloud platform isn't automatically the right answer. The solution should follow the company objectives, not the other way around.

Start with an AI pilot to prove business value before scaling

AI adoption doesn’t need to start with an enterprise-wide initiative. (In fact, it shouldn’t).

We suggest starting with a minimum viable product (MVP) tied to a clear business use case. Test the concept with a small team, validate the approach, and measure the results before expanding.

You might also find that existing tools (such as Copilot) are enough to meet your current needs. This is the perfect space to begin by testing with a specific team and carefully measuring results. You can always add more customized solutions as requirements grow. The goal isn't to build the smallest solution indefinitely; it's to avoid making large investments before the business case has been demonstrated. Why dedicate precious resources before you need to?

How to evaluate AI use cases: Business value, feasibility, cost and risk

A promising use case also needs to be technically feasible, economically justified, and appropriate for its level of risk. During the webinar, BBI Chief Technologist Vivek Gopinathlal outlined the BTER framework:

  • Business value: What measurable outcome will the solution deliver?
  • Technical fitness: Can the existing environment and data support it?
  • Economics: Do the expected benefits justify implementation and operating costs?
  • Risk: What controls, governance, and human oversight will be required?

This framework can help you to decide which opportunities to pursue, which need further assessment, and which may not be worth the investment at all. You might be surprised at what you find.

Identify high-value Internal AI use cases

As mentioned, internal applications can be a useful starting point for AI, particularly when the goal is to improve employee productivity or make existing information easier to use.

For example, retrieval-augmented generation (RAG) can help employees find answers across internal documents and repositories such as SharePoint or Confluence. Other opportunities include standardizing software development documentation, improving access to institutional knowledge, and surfacing project information about schedules, resources, and budgets.

The best use cases depend on the organization's needs and the reliability of its source material. Internal applications can also provide a more controlled environment for testing before moving to customer-facing solutions.

Why data quality and governance matter for AI readiness

AI can work with documents and other unstructured information, but its ability to process data doesn't guarantee reliable answers. Outdated, inconsistent, or incomplete source material can undermine results. The old adage still stands: garbage in, garbage out.

Organizations need to assess data quality, establish governance and access controls, and evaluate outputs. Importantly, it’s a good idea to make answers traceable to their sources where possible. For consequential decisions, human accountability is still essential. AI can flag potentially fraudulent transactions for review, but a person should retain approval authority where the decision requires it.

Over time you can expect AI to take on more responsibility in your company – but the more heavy lifting AI systems do, the more carefully controls and decision boundaries must be designed.

Why cloud migration alone isn’t enough for AI readiness

Moving systems to the cloud doesn't automatically make them more capable or valuable. A migration can preserve existing inefficiencies while introducing additional costs for data movement, integration, skills, and operations.

Organizations should assess which legacy systems and data remain useful, what needs to change, and what isn't worth carrying forward. You aren’t trying to simply recreate an existing environment in the cloud. You want to create an environment that supports more reliable data, better use of information, and future business requirements. Otherwise, what’s the point?

What should come first in AI readiness: Business priorities, data or technology?

Ravneet Singh, BBI CEO, has spent a lot of time thinking about this, and offers a useful hierarchy for evaluating AI investment:

Data matters more than automation.
Foundation matters more than data.
Picking the right task matters more than foundation.

Each layer matters, but the order is important. Automation built on unreliable data can reproduce errors faster. Data without a sound technical foundation may be difficult to use effectively. And even a strong foundation has limited value if it is built to solve the wrong problem.

The right task gives the investment direction. From there, organizations can determine what data they need, which technical capabilities are required, and what level of automation makes sense.

Build an AI modernization strategy around business value

There is no single platform or AI use case that is right for every department or every organization. The path forward depends on the desired business outcome, the existing environment, data quality, economics, and risk.

At BBI, we believe that AI readiness isn't a race to simply adopt more technology. It's about building solutions that address real business needs and bring benefits in the real world.